QUANTIFYING BIOSTRATIGRAPHIC UNCERTAINTY FOR BAYESIAN TIMESCALE CALIBRATION

Numerical calibration of the Paleozoic portion of the geological timescale is typically accomplished by fitting a linear regression, LOWESS, or cubic spline curve to radioisotopic dates pinned to some stratigraphic interval. The uncertainty in the radioisotopic date is a well-known issue that encompasses the propagated analytical, tracer, and decay-constant uncertainties of the determined numerical age of the sample. The uncertainty in the stratigraphic position of the age date, however, has been a largely overlooked factor in the uncertainty of timescale calibration overall. What limited evaluations have been done demonstrate that the temporal uncertainty associated with the stratigraphic position of the dated bed is typically far greater than the uncertainty associated with the radioisotopic date itself.

Multiple methods for quantitative stratigraphy have been developed in the past 50 years, including powerful computer programs such as CONOP or Horizon Annealing. Each of these is capable of providing methods for determining uncertainty estimates for biostratigraphic data, including the first and last appearances of species in a given stratigraphic section. However, these uncertainties are typically not presented as probability distributions of uncertainty in the same way that we typically see radioisotope data. With the recent progress in Bayesian methods of timescale calibration, the ability to determine probability density as a measure of stratigraphic uncertainty will become an increasingly critical issue and should be a new primary objective of the biostratigraphic community. The ability (or inability) to determine the temporal consequences of paleobiogeographic diachroneity due to origination, distribution, and refugia, has now become the largest uncertainty in the calibration of the geologic timescale and represents a major new objective for the paleontological community going forward.